Team identification as a key variable to distinguish supporters of a football club fan base

(1)

Team identification as a key variable to distinguish

supporters of a football club fan base

By

Michiel van Gessel

(2)

Title: Team identification as a key variable to distinguish supporters of a football club fan base

Author: Michiel van Gessel

Department: Marketing

Qualification: Master thesis

Completion date: March 24, 2011

(3)

Management Summary

The role of marketing is increasingly claiming its importance in the sports industry, as the management of sports organizations are more and more reckoning its usefulness. The treatment of one’s fan base, which is the backbone of a club’s existence, is especially receiving attention among both academics and practitioners. Furthermore, to counteract against the unpredictability of success in team sports, management should diminish their reliability on on-pitch performances and adopt additional off-pitch strategies, which focus on a long-term and steady revenue stream (Kaynak, Salman, and Tatoglu 2007).

Although most football clubs currently concentrate on transaction marketing, relational activities are more capable of provoking customer loyalty to ensure a stable, mutually profitable, and long-term relationship (Ravald and Gronroös 1996). In addition, Customer Relationship Management (CRM) applications can provide a competitive advantage in leveraging their collection of customer data to customize offerings and respond to differing customer needs (Mithas, Krishnan, and Fornell 2005).

This study, which is executed in cooperation with a well-known Dutch football organization, conceptualizes the relationships between supporter demographics, attitudes and behaviours. Whereas a supporter database was used to obtain supporter demographics, a web-based survey was executed to uncover latent attitudes and behaviours that supporters perform. As research on fan behaviour has revealed that the major difference among sports fans and spectators affecting their consumer behaviour is that of team identification (Shilbury et al. 2009), the influence on and consequences of team identification are the primary focus of this research. With the use of a mediation analysis this study confirms the importance of team identification, as it significantly intervenes or mediates the influence of both age and gender on all types of behavioural outcomes presented in the conceptual model. This yields both insights in the underlying motives that result in certain fan behaviours and clarifies the importance to monitor team identification throughout the fan base.

(4)

levels of team identification than older ones. However, no evidence was found that supporters who live in the near proximity of a sports organization possess higher levels of team identification than supporters who reside in more distant locations. Moreover, when focusing of the attitude-behaviour relationships, team identification has a significant and positive relationship with relationship length, loyalty intentions, and word-of-mouth. As these behaviours are all beneficial to a sports organization, management should focus on increasing the level of team identification. In addition, perceived quality only seems to moderate the latter relationship, indicating that fans with high levels of team identification are more likely to express favourable word-of-mouth when they are accompanied with a high perceived quality.

(5)

Preface

Although winning in sports is the ultimate goal, the journey of dedication and motivation towards success is fundamental for every athlete. After being a scholar for almost two decades, this phrase seems appropriate for both sport and educational platforms. This master thesis is my concluding report, demonstrating the knowledge that I gained during my bachelor and master studies. I would like to take the opportunity to thank a number of people who have contributed to the conclusion of this master thesis.

First of all I would like to thank prof. dr. Janny Hoekstra, who supported and provided me with feedback throughout the production of my thesis. Even though sport marketing is a small topic within the Marketing masters at the University of Groningen, she was enthusiastic about my ideas right from the start. Furthermore, dr. Erjen van Nierop, who gave a critical look at my results analyses, was also really supportive in the process of fine-tuning the analyses of my thesis.

My parents have always been my greatest supporters, I am very grateful that you provided me the opportunities to realise my goals. You have always believed in me and I am truly aware that it is a blessing to have good and caring parents. I would not have achieved this result without you. Furthermore, both my brothers and sister have been supportive all the way, and provided me with their knowledge and experience gained during studies and practice. Although not on my side during the entire process, Annya Klamer has been of great support during the initiation process of my thesis. You really stimulated me to produce a thesis that combined study and passion. Thank you for everything.

Finally, I would also like to thank Martin van der Meulen, who gave me the opportunity to work with a professional sports organization, and who granted me access to a supporter database. In addition, Ellis Netjes and Robert Dijkstra were really helpful during the data collection process. Thank you all.

(6)

1 Introduction and problem specification

1.1 Introduction

Nowadays, the football industry is showing more and more resemblances with traditional businesses. The constantly growing turnovers of football clubs have made this industry an interesting area of research in general. The field of marketing has also made its entrance in the world of football, as the management of football teams are becoming more concerned with the treatment of their fan base. Whereas a club’s fan base accounts for the backbone of its existence, the treatment of these fans should be dealt with care.

As with businesses in almost every sector, the goal of management is to expand the revenue stream resulting from business operations. The sports industry however is characterized by several aspects that diminish the opportunities to enlarge revenues compared to other industries. As is the case with many services, the main problem is that the supply of the core sport product – the on-field performance – cannot be increased in the same way as a manufactured good. In addition, sport clubs and competitions are traditionally restricted to what economists would call a fixed short-run supply, or a highly inelastic production curve (Stewart and Smith 1999). Focusing primarily on short run and on-pitch performances, such as signing star players and raising salaries, has not proven to be a flawless strategy for many clubs. This is because it does not consider the unpredictability of success in team sports. To counteract against these short term tactics, management should therefore adopt additional off-pitch strategies, which focus on a long-term and steady revenue stream (Kaynak, Salman, and Tatoglu 2007). As sport is still characterized by fierce, loyal and passionate fans who experience a strong, vicarious identification with their favourite players and teams (Smith and Stewart 2010), the management should try to exploit possibilities that result from different fan preferences.

(9)

and effectiveness of the interaction to attract and satisfy consumers (Agustin and Singh 2005). However, the benefits of relational activities are often studied by researchers. They find that the use of Customer Relationship Management (CRM) applications is positively associated with improved customer knowledge and improved customer satisfaction (Mithas, Krishnan, and Fornell 2005). Consequently, it provides a competitive advantage in leveraging their collection of customer data to customize offerings and respond to customer needs (Mithas, Krishnan, and Fornell 2005). For instance, offering augmented products or services to certain fans could be used for cross subsidizing the on-field activities of sport organizations (Smith and Stewart, 2010).

Research on fan behaviour has revealed that the major difference among sports fans and spectators affecting their consumer behaviour is that of team identification (Shilbury et al. 2009). Consumers exhibiting high levels of identification will potentially benefit organizations with increased loyalty and positive word-of-mouth (Bhattachary and Sen 2003). They argue that the identification of these consumers with the use of CRM applications could result in favourable outcomes, as they can be targeted and treated according to their needs. Furthermore, it enables the design and development of better products and services (Davenport, Harris, and Kohli 2001; Nambisan 2002).

(10)

1.2 Research questions

According to Shilbury et al. (2009), the major differential among sport fans is their level of team identification. As team identification is a latent attitude that sport fans posses, this information is difficult to retrieve for organizations. However, marketing managers are very interested in the extent to which sport fans indentify themselves with a certain team, as these people tend to behave differently. Attitudes, such as team identification, are also of interest as they are often guiding these behaviours. This attitude-behaviour relationship is important to assess for marketers, as they might be able to predict actions (Hoyer and MacInnes 2007), or respond to them.

In addition, sport fans deploying a specific level of team identification might also posses certain demographic characteristics. As demographic variables do not prove to be very sensitive and are often stored and monitored by sport organizations, the relationship between demographic variables and team identification can be assessed.

As stated previously, the degree of team identification might also have an influence on the behaviours that supporters deploy. This might be of relevance to a sport organization, as this attitude might provoke positive behaviours of supporters and their surroundings. Identifying, and consequently treating these supporters with customized offerings might offer sport organizations a competitive advantage.

Furthermore, it might be the case that supporter demographics exert influence on supporter behaviours through intervening or mediator variables such as team identification. Assessing the mediating effect of team identification yields insights in the underlying motives that result in certain fan behaviours.

Finally, this study considers the role that market segmentation potentially could have in guiding a sport marketing strategy. This to derive the value that a football club can obtain when maintaining a database of their supporters. The aim of such strategy is to move each segment closer to loyalty by increasing their level of team identification.

(11)

- To what extent is it useful to differentiate supporters according to their levels of team identification?

- Which demographic variables influence the level of team identification?

- How are behaviours of supporters influenced by the level of team identification? - To what extent is team identification mediating the relationship between

demographics and behaviours of supporters?

- Which segments can be distinguished based on the demographics, attitudes and behaviours that sport fans possess?

1.3 Theoretical and managerial relevance

There is an increasing body of literature that discusses marketing strategies for sports clubs nowadays. Furthermore, attitudes such as team identification increasingly attract the attention of academics (eg. Matsuoka, Chelladurai, and Harada 2003; Mullin, Hardy, and Sutton 2007; Smith and Stewart 2010; Wann et al. 2001; Wann and Branscombe 1993). This literature emphasizes the need to reflect the uniqueness of the sports spectator, and the implication is that consumers in this sector should be treated as different to those of arguably more “mainstream” areas of commerce (Tapp and Clowes 2002).In addition, the potential benefits of CRM application within the sports industry has been studied to some extent (Adamson, Jones, and Tapp 2005), and practitioners are more and more reckoning their usefulness.

(12)

In addition, clubs could benefit from this research as the proper identification of fans exhibiting different levels of team identification could potentially result in a competitive advantage. As it, for instance, offers clubs the opportunity to customize their offerings to certain supporter segments.

1.4 Structure of the thesis

The aim of this research study is to establish certain relations between demographics, attitudes and behaviours of football supporters. In addition, this study also assesses the value that clubs can derive from embedding CRM application within their strategy, as this can be a tool for identifying relations between variables. Chapter 2 presents a review of prior research concerning the business of sports as a whole, CRM strategies in general, behaviours of fans, and advantages of clustering a database into segments. Furthermore, current trends within European football and their consequences for football clubs are evaluated in this chapter. Subsequently, a theoretical framework, including hypotheses, that is built upon a conceptual model will be presented in chapter 3. The research design, data collection, and plan of analysis will be discussed in chapter 4.

The results of this study will be presented in chapter 5. This chapter kicks off with some descriptive statistics that give an overview of the population sample that has been used. Furthermore, the relations and hypotheses presented in chapter 3 are tested on their directions and significance. In addition, a cluster analysis is performed to assess the segmentation possibilities that can be derived when managing the variables discussed in the theoretical framework. This in order to obtain insights in the possible gains that result from employing different marketing tactics for separate segments opposed to providing one broad strategy for the entire fan base. Furthermore, the mediating role of team identification in the relationships between supporters demographics and behaviours is assessed with the use of mediation analysis.

(13)

2 Literature Study

As this study focuses primarily on the sports industry, the major differences and similarities with other businesses are mentioned first in section 2.1. Furthermore, some of the major trends in the European football competitions are described in section 2.2. These trends concern both revenue extracting sources as well as managerial issues that are gaining popularity. In addition, customer relationship management (CRM) is introduced in section 2.3, which already has proven its value in many other businesses. The potential gains of such practices are mentioned, and the suitability for such a tool in the football industry is evaluated. Finally, the advantages of clustering a customer database are presented in section 2.4. Especially within in the sports industry, where such practices are not that common, it is interesting to asses its potential value.

2.1 Characteristics of the sports industry

When comparing professional sports to other businesses, multiple distinctions can be addressed that outline the uniqueness of the sports industry. According to Mullin, Hardy, and Sutton (2007), and Shilbury et al. (2009), the sport market differentiates itself by special characteristics. A number of these characteristics are:

- Many sport organizations simultaneously compete and cooperate in the same marketplace;

- Product salience and strong personal identification lead many sport consumers to consider themselves as experts;

- Sport evokes powerful personal identification and emotional attachment;

- Sport has almost universal appeal and pervades all elements of life, i.e, geographically, demographically and socioculturally;

(14)

- Sport can be seen as a social event – consumers of sports watch sports together and discuss sports in daily life and

- It is hard to control the core (game) of the product.

Thus, marketers within the sports industry are facing different challenges than other industries when marketing their products. Therefore they need to build upon the understanding of traditional marketing (Funk 2008).

However, the management of sport has developed into a delicate issue recent years. This because the management of sport has been divided into two contrasting approaches

(Stewart and Smith 1999).At one extreme, sport is viewed as a unique cultural institution with a host of special features wherein the application of standard business practices not only produces poor management decision making, but also erodes its rich history, emotional connections, tribal links, and social relevance. At the other extreme, sport is deemed to be nothing more than just another generic business enterprise subject to the usual government regulations, market pressures and customer demands, and is best managed by the application of standard business tools that assist the planning, finance, human resource management and marketing functions (Stewart and Smith 1999).

Other issues that could create tension within a sport organization are the different views that management and fans possess. Whereas the fan base of a sports club is primarily interested in on-pitch results and cultural aspects, shareholders are keen on the value of the club, which in turn is a direct result from on-pitch performance. As a consequence, fans and management share their desire for good on-pitch performance, but differ in their attitudes towards culture and shareholder value. However, as the product – on pitch performance - is inconsistent and unpredictable, management has less control over the core product than other ‘normal’ businesses (Mullin, Hardy, and Sutton 2007).

(15)

certain number of games can be played during a season, and irrespective of the spectator demand, attendance is always limited by the number of games scheduled and the seating capacity of the venue. Conversely, when there is limited demand, unsold seats represent revenue lost forever. In these instances the sport product cannot be stored and re-sold another day.

In addition, even though the management of a sport organization has no control over the core product, they have the responsibility of marketing all related activities to a demanding and heterogeneous fan base. Furthermore, sport is a social institution which also needs to act as a business and should attract commercial value. This paradox of commercialism also separates the interests of supporters and stakeholders, who both should be accounted for by the management.

2.2 Trends in the (European) Football industry

(16)

Besides the domestic league, participation and relative performance in the UEFA Champions League continues to be a key factor in determining a club’s revenues. This because in addition to the centrally generated revenue, clubs directly generate matchday revenue and, indirectly, their participation also boosts their sponsorship and other commercial revenues both in the short term, through contractual bonuses, and longer term by strengthening their brand attractiveness through increased exposure and profile. For a successful club in a large market, this can add up to a highly significant revenue stream (Deloitte and Touche 2010). In addition, it seems correct to state that better on pitch performance results in better financial performances. In turn, better financial performances offer the clubs the potential to attract star players, which once again could result in better on pitch performances. This vicious circle is very attractive for many clubs within Europe, and they often spend huge amounts of resources on realizing this goal. Given the potential rewards just mentioned it is perhaps not surprising that some clubs commit to certain costs (principally in terms of player salaries) in the hope, and expectation, of achieving Champions League football and the additional revenues. However, crafting a strategy based on alleged on-pitch successes has not proven to be very effective in the sports industry, where multiple uncontrollable (endogenous) elements have an effect on performance.

The primary goal for the management of football clubs in general should be that they diminish their reliability on on-pitch performances. In addition, the board should also focus on expanding the revenue stream resulting from off-pitch activities. This should result in a strategy with a long-term focus considering the unpredictability of success in team sport. Short-term tactics are commonly used by sport managers such as firing a head coach, or signing a new player. No matter how important winning may be, these short-term tactics do not guarantee long-short-term and steady revenue flows. Sport managers, therefore, should adopt brand management strategies, realizing that winning is only one aspect of consumers’ experience (Kaynak, Salman, and Tatoglu 2007).

(17)

cross-subsidizing the core on-field activities of sport organizations, they have to be handled with a careful awareness of customer satisfaction and service quality (Smith and Stewart 2010).

Nowadays, the sports industry faces new challenges, as fans have more leisure choices and are more discerning in terms of their service expectations. Clubs have become more commercially focused and are beginning to show signs of developing conventional business approaches such as relationship marketing. Traditionally, it has been difficult for clubs to maintain contact with their fan bases, but the increased popularity of new technologies, such as the Internet in general and social media in particular, have resulted in new opportunities to communicate with fans (Adamson, Jones, and Tapp 2005).

2.3 Customer Relationship Management

(18)

The benefits of such activities are often assessed by researchers. For instance, Mithas, Krishnan, and Fornell (2005) found that the use of CRM applications is positively associated with improved customer knowledge and improved customer satisfaction. Furthermore, they found that gains in customer knowledge are enhanced when firms share their customer-related information with their supply chain partners. Moreover, Reinartz, Krafft, and Hoyer (2004) found that the implementation of CRM processes is associated with better company performances in the initiation and maintenance stage. However, although there appears to be general consensus on the importance of CRM as a strategic imperative among both academics and managers, the return on investments in CRM strategy and programs seem to vary, both within and across organizations (Bohling et al. 2006).

A primary motivation for any type of firm to implement CRM applications is to track customer behaviour to gain insights into customer tastes and evolving needs (Mithas, Krishnan, and Fornell 2005). By organizing and using this information, firms can design and develop better products and services (Davenport, Harris, and Kohli 2001; Nambisan 2002). Furthermore, firms with a greater deployment of CRM applications are likely to be more familiar with the data management issues involved in initiating, maintaining, and terminating a customer relationship. This familiarity gives firms a competitive advantage in leveraging their collection of customer data to customize offerings and respond to customer needs (Mithas, Krishnan, and Fornell 2005). To sustain success in the current market place, more and more companies are attempting to build deep, meaningful, and long-term relationships with their customers (Bhattacharya & Sen 2003). The ultimate success for a company is that in addition to its loyalty, the consumer champion also enthusiastically promotes the company and its products to others (Bhattacharya & Sen 2003). The challenge for a company is to deliver its brand promise in such a way that everyone really appreciates what the company stands for and feels committed to the company (Chernatony 2006).

(19)

(2005) state that the football industry is lagging behind conventional businesses in employing CRM; rather than enjoying the benefits of being a ‘follower’ — in terms of the opportunity to learn from other industries — it gives the appearance of starting from scratch instead. Focusing on the football industry, transaction marketing seems to be more appropriate according to many clubs. This transactional view of B2C relationships emphasizes the one-time provision of economic benefit, profit, efficiency, and effectiveness of the interaction to attract and satisfy consumers (Agustin and Singh 2005).

However, Ravald and Gronroös (1996) argue that the focus of marketers should shift from the activity of attracting customers to activities which focus on maintaining customers and taking care of them. They state that it is important to create customer loyalty to ensure a stable, mutually profitable and long-term relationship. In this light, relational activities have gained popularity within the sports industry. Moreover, the escalator of Mullin, Hardy, and Sutton (2007) states that relational activities are more appropriate than transactional, suggesting that sport organizations should invest more in nurturing existing consumers than they should in trying to create new ones. In addition, Tapp’s (2004) work suggests that for some fans a relationship building approach may be more appropriate in supporter retention, while for other types of fans transaction marketing may be profitable. Hence, different kind of supporters ask for different kind of treatments. A segmented approach enabled by CRM could pay dividends.

(20)

According to Adamson, Jones, and Tapp (2005), loyalty levels are assumed to be much greater within sports than in conventional industries. They conclude that marketers of football clubs believe that they can take their fans’ loyalty for granted. However, marketers of football clubs should be aware that the sports industry has some unique characteristics (Please refer to section 2.1 for a detailed description) and in some situations fans should not be treated as a company consumer. They state that there are different segments of football fans and that marketers should look into different approaches to build a relationship with a fan (Adamson, Jones, and Tapp 2005).

However, despite the managerial drive in sport for more revenue and improved efficiency, many sport fans still argue for the prioritization of on-field success and the celebration of competitive ideals which privileges it above conventional profit seeking endeavours (Smith and Stewart 2010). This could pose a problem when a club is trying to implement CRM applications, as some fans are reluctant to respond to such instruments.

2.4 Segmentation

(21)

newsletter, and a stadium tour. Segmentation is a fundamental tactic of marketing designed to maximize consumer satisfaction and demand, according to their identified needs (Shilbury et al. 2009).

For sport marketing decision-makers to use this information to develop marketing strategies requires an evaluation of how their service offering matches the market. The segmentation process must define the market and the manner in which it is to be satisfied. A season ticket holder is more valuable than a one-time TV viewer, and should be considered as such. Therefore the importance of segmentation in informing sport marketing decision-making and strategy formation is critical (Shilbury et al. 2009).

According to Adamson, Jones, and Tapp (2005), CRM practices are very well capable of enabling a segmentation approach. In turn, marketing decision-makers need to appreciate the importance of identification levels among customer segments. Influencing the degree of identification and loyalty is a critical objective that needs to be part of sport marketing strategy and decision-making. The aim of such strategy is to move each segment closer to loyalty by increasing their level of identification with the sport itself, the club and players or product extensions integral to the sport entertainment package offering. Achieving this objective is necessary to make the sport more financially attractive for the sport property, the media and sponsors alike (Shilbury et al. 2009).

(22)

(23)

3 Conceptual Model and Hypotheses

In order to develop a thorough understanding of their fan base, a football club needs to monitor and keep track of the characteristics, attitudes, and behaviours of their fans. In addition, the relationships between these variables can yield very interesting insights for the management of a football club. As team identification is the major differential among sport fans, this will be elaborated first in section 3.1. Second, the antecedents of team identification and their relationship with team identification are discussed in section 3.2. Third, section 3.3 elaborates the consequences of team identification, which comprehends three types of behaviours that fans usually deploy. Finally, the moderating influence of perceived quality on the attitude-behaviour link will be assessed in section 3.4.

3.1 Team Identification

Research in fan behaviour, an element of sport marketing, has revealed that the major differential among sports fans and spectators affecting their consumer behaviour is team identification (Shilbury et al. 2009). Team identification signifies the extent to which a fan feels a psychological connection to a team, is involved and interested in the team, and sees the team as an extension of the self (Wann, Bayens, and Driver 2004; Wann and Branscombe 1993). In addition, team identification can be seen as an attitude, which is an overall evaluation that expresses how much we like or dislike an object, issue, person or action (Hoyer and MacInnes 2007). The level of identification a fan has might be referred to as a level of psychological attachment, which can correspond to varying levels of loyalty being manifested. Wann et al. (2001) argued that such a psychological attachment is stable and does not fluctuate from season to season, or after a win or loss. This makes team identification an excellent variable for this study.

(24)

greater desire to affect the result of an event, exhibit great anxiety while watching an event, enjoy spectatorship more than low team-identified fans, possess greater team knowledge, and team success may be highly connected to their sense of self and self esteem (Shilbury et al. 2009). An example of high identification within a family may be where they often discuss sport, frequently watch sport, subscribe to several sports magazines, take family trips to sporting events, may have a team membership and may interact with other fans over the internet on team websites or sports based websites. In contrast, low team identification is associated with fans where their team is peripheral to their life and sense of self, they are less connected to their team, and they are less likely to enjoy spectatorship (Shilbury et al. 2009). An example of low team identification may occur where a corporate guest is invited by a company employee to watch a football match. The guest may never or rarely attend games and may only occasionally take note of related sport media, with the reason for attendance having more to do with building business relationships rather than a high level interest in a team or the sport.

Marketing decision-makers need to appreciate the importance of identification levels among customer segments. Influencing the degree of identification and loyalty is a critical objective that needs to be part of sport marketing strategy and decision-making (Shilbury et al. 2009). The aim of such strategy is to move each segment closer to loyalty by increasing their level of identification with the club or product extensions integral to the sport entertainment package offering. Achieving this objective is necessary to make the sport more financially attractive for the sport industry, the media and sponsors alike.

(25)

of a company depends on the extent to which consumers perceive it to be similar to their own identity, the dimensions they value, and the prestige of the company. Furthermore, the strength of identification with the company depends on the level of knowledge the consumer has about the company identity. Moreover, consumer-company identification is influenced by the company’s ability to present its identity in a coherent and internally consistent way (Bhattachary and Sen 2003).

When comparing the consumer-company relationship with a consumer-sports club relationship, Wann (1997) states that a relationship is determined by the consumer’s psychological connection and attachment to a team. This implies that a consumer with strong team identification considers the sports club as a central component of its self-identity. In addition, consumers with low levels of team identification do not strongly relate the sports club to their self-identity.

From a sport marketing management perspective the aim is to understand the motives and needs that exist at various levels of team identification (Mullin, Hardy, and Sutton 2007). Having this knowledge will inform sport-marketing strategists who seek to take less loyal (low identification) fans up the loyalty escalator to become more highly identified, loyalty-committed fans.

(26)

Finally, it can be stated that team identification has an overlap with commitment as both variables refer to a level of attachment with a certain organization. In addition, according to Mowdow, Porter, and Steers (1982), commitment is seen as the relative strength of an individual’s identification with a particular organization.

3.2 Antecedents of Team Identification: demographics

Although the role of demographics and personality characteristics has often been assessed by researchers, their relationship with team identification has not been of primary interest. Wann et al. (2001) provides a literature overview of demographics and personality characteristics associated with sport fandom. As an overlap might be expected between sport fandom and team identification, this literature overview can be used to estimate the relationship between demographics and team identification. Focusing on gender, a disproportionately share is likely to be male according to Wann et al. (2001). However, this does not suggest that males exhibit higher levels of team identification than females. Findings of Wann and Branscombe (1993) support this statement as they did not find any differences in gender for different levels of team identification. Furthermore, Robinson and Trail (2005) anticipated that if there were differences in the level of team identification for gender, the differences would not be large or meaningful. However, results of Gwinner and Swanson (2003) indicate that highly identified fans were more likely to be male. Even though previous studies do not coincide with the relationship between gender and level of team identification, the following hypothesis is formulated.

H1. Males and females do not significantly differ in their levels of team identification.

(27)

H2. There is a positive relationship between age and the degree of team identification.

When referring to the relationship between residence and team identification, it can be stated that local fans are more likely to exhibit fan-like behaviour than displaced fans because of identification with a geographic area (e.g. where he or she was born or lives) (Hunt, Bristol, and Bashaw 1999). In addition, Jones (1997) found that geographical reasons were by far the most prominent ones listed for identification with a favourite team. His findings suggest that the two most frequently cited reasons by fans for currently supporting their favourite football team were that it was the local team and that the fan was born in the town or city. Furthermore, another common reason for originally identifying with a team was that one’s parents were supporters of the team. Moreover, Wann et al. (2001) noticed that a complex interaction exists among factors of geography, team success and team identification. Their study revealed that team success did not have a significant impact on team identification when comparing all fans of a team, yet when geographically displaced fans alone were measured, team success had a high correlation with team identification. Therefore, Wann et al. (2001) concludes that team success is more closely related to the team identification of displaced fans than of local fans. This is of interest as it suggests that geography alone is one form of segmentation among fans of a sporting club. As a consequence, the degree of team identification from fans who life nearby are less influenced by on pitch performances than fans who reside further away. This results in the following hypothesis.

H3. Fans who live in or nearby the hometown of a particular football team exhibit higher levels of team identification than fans who live in more displaced areas.

3.3 Consequences of Team Identification

(28)

3.3.1 Relationship Length

As stated previously, Wann and Branscombe (1993) found that individuals high in identification with a sports team are more involved with that team. This involvement is also translated into a greater number of years as a fan, which in turn indicates a longer relationship length. Prior research (Bolton, Lemon, and Verhoef 2004) also suggests that there may be positive effects of commitment on relationship duration. Whereas commitment shows some overlaps (see section 3.1) with team identification, these findings can also be applied for team identification. In addition, it can be stated that team identification has a positive influence on relationship length.

Furthermore, retention can be seen as a strong determinant of relationship length as people who are retained also expand the length of their relationship. Research of Gustafsson et al. (2005) suggests several predictors of retention. First, affective commitment has an influence on retention as it captures the trust and reciprocity in a relationship. Second, calculative commitment also proves to have a relation with retention as it captures the existence of switching costs or lack of viable alternatives (Gustafsson et al. 2005). Consequently, as commitment shows some overlap with team identification, it can be stated that team identification has an influence on the relationship length.

H4. There is a positive relationship between the degree of team identification and the duration of the relationship between sport organization and supporter.

3.3.2 Word-of-Mouth

(29)

The impact of word-of-mouth (WOM) on consumers’ actions, preferences, and choices has been of great academic interest recent years. Many researchers have already corroborated the primacy of WOM as a key driver of firm sales (Godes and Mayzlin 2009). In addition, the link between identification levels and WOM has also received attention in several studies. For instance, Bhattachary and Sen (2003) found that consumers who experience strong identification with a company will benefit the organization as the consumer will:

- Be more loyal to the company - Promote the company

- Recruit customers actively

- Be resilient to negative information - Claim stronger on the company

In addition, Westbrook (1987) pointed out that consumers are more likely to engage in word-of-mouth when they experience notable emotional experiences. Consumers who are most likely to experience such emotions are probably fans with high levels of team identification. Furthermore, observations have shown that most supporters come in groups either with friends or with family members (Decrop and Derbaix 2010), who all serve as personal information sources. As the urge for social contacts is one of fans’ most basic drives, it might be expected that supporters with higher levels of team identification are more reluctant to express positive communications about their favourite team. In addition, the following hypothesis is presented.

H5. There is a positive relationship between the degree of team identification and the extent of personal information expressed and shared (WOM).

3.3.3 Loyalty Intentions

(30)

and Howard 2000; Trail, Fink, and Anderson 2003; Wann et al. 2001). When referring to loyalty intentions, Johnson, Herrmann, and Huber (2006) found that the drivers of these intentions are dynamic and their influence changes over time. Whereas perceived value is the main driver influencing loyalty intentions at early stages of the relationship, affective attitudes such as team identification drive intentions as the relationship continues (Johnson, Herrmann, and Huber 2006).

In addition, Matsuoka, Chelladurai, and Harada (2003) found that team identification is significantly correlated with someone’s intention to attend future games. Similarly, Trail, Fink, and Anderson (2003) noticed that team identification was a key factor in explaining future consumption decisions of spectators at intercollegiate basketball matches. Furthermore, Tobar (2006) studied television spectators of the Super bowl XL and established that stronger affiliations were related to higher scores for purchase intentions. Likewise, Tsiotsou and Alexandris (2008) found that team attachment, which can be seen as a synonym for team identification, exhibited the strongest positive total effect on purchase intentions. Moreover, Wann and Branscombe (1993) found that individuals with high levels of team identification exerted greater attendance in both home and away games as well as high expectations for future attendance-related behaviours. Finally, Madrigal (2000) showed that fans of college football purchased more merchandise as their identification with the team increased. Consequently, the following hypothesis can be derived from the various studies mentioned above.

H6. There is a positive relationship between the degree of team identification and loyalty intentions.

3.4 Moderators of Team Identification

(31)

attendance, poor team performance and game outcomes lead to dissatisfaction which, in turn, results in decreasing attendance. Although the outcomes of a football match do not entirely depend on the quality of the players on the pitch, it definitely has an influence on the chances of winning. Whereas the actual quality of a team is almost impossible to measure, the perceived quality or value that a supporter attaches to his or her favourite team can be derived. Zeithaml (1988) has suggested that perceived value can be regarded as a consumer’s overall assessment of the utility of a product or service based on perceptions of what is received and what is given. Furthermore, it can be stated that factors influencing perceived value, such as perceived quality and on-pitch results of a certain team, guide someone’s intentions and might possibly turn a general sport fan into a specific sports team fan (Johnson, Herrmann, and Huber 2006).

In addition, perceived quality could potentially moderate the attitude-behaviour link. For instance, the game play and outcomes of a certain team might have influence on the perceptions of fans. Especially since perceived value is ‘backward looking’, extreme values could potentially moderate the ‘forward looking’ attitude-behaviour link (Gustafsson, Johnson, and Roos 2005). For example, fans with high levels of team identification might be reluctant to express word-of-mouth concerning their favourite team, as recent matches have been boring and the results below expectation.

H7. Perceived quality moderates the effect of team identification on the behaviours of fans.

3.5 Team Identification as a mediator

(32)

supporter demographics on behavioural supporter outcomes. Hoyer and MacInnes (2007) already stated that attitudes have a direct influence on behaviours. For instance, in relation with the current study, sport marketing research (see section 3.3) revealed that sport fans with varying levels of team identification tend to behave differently. However, as there is no support that supporter demographics have a direct effect on behavioural outcomes, it is suggested that team identification mediates the influence of supporter demographics on behavioural supporter outcomes. Although not conceptualized in figure 3.1, the following hypothesis is presented.

H8. Team identification mediates the causal sequences between the demographics and behaviours of supporters.

Figure 3.1 summarizes all the variables and their inter-relationships mentioned in this chapter. This framework is a conceptual model indicating how demographics influence the level of team identification that a supporter has, and shows the behavioural consequences that result from certain levels of team identification.

(33)

4 Research Design

The following chapter provides a detailed description of the research design that is developed to guide the research study towards its objectives.

4.1 Research method

In order to test the relationships presented in the conceptual model, data from multiple sources are collected. The collection method and the number of respondents will be mentioned in section 4.2. Furthermore, as several multi-item scales are allocated in the web-based survey, underlying theory and actual results are presented in section 4.3. Finally, the most important analysis methods applied (regression, latent class and mediation) in this study are discussed in section 4.4.

4.2 Data Collection

A supporters database will be analyzed in order to asses the opportunities that can be derived from such a database when executing CRM practices. By the use of a web-based survey several subjective variables (such as team identification, WOM, and loyalty intentions) will be combined with demographic data from the database in order to assess relations between characteristics of fans and these variables.

(34)

objectives of the study were clearly mentioned to inform the supporters of the potential gains that could be derived for both the club as themselves. Second, the estimated time that was needed to properly fill in the questionnaire was mentioned, so it would be clear that the club only asked for a small favour. Finally, an official match day jersey, signed by all team and staff members, would be raffled among the respondents. Appendix 1 shows the actual questionnaire that was used in this study. The main part of the questionnaire contains of statements, all accompanied with a 7 point likert-scale to assess the attitudes and behaviours of the participants. Furthermore, open questions were used to derive the relationship length and perceived quality. The web-based survey resulted in 331 respondents from a total sample of 1200 supporters. This indicates that the response ratio was just under 30%. When applying a statistical formula1, the sample (331) could be designated as representative when it is accompanied with a confidence interval of 95% and a confidence level of 5.36%. In addition, as all issues concerned with data collection are addressed adequately, the proportion in this study is set at 50%.

4.3 Measurement Scales

The internal consistency of multi-item scales needs to be assessed in every research setting. Although previous studies suggested adequate internal consistency for both the team identification and word-of-mouth construct, reassessing these results strengthens the validity of the study. To evaluate the acceptable fit for both the constructs, goodness-of-fit statistics are used such as Cronbach’s alpha. The generally agreed upon lower limit for Cronbach’s alpha is 0.70 (Malhotra 2007). One issue in assessing Cronbach’s alpha is its positive relationship to the number of items in the scale. An increased number of items will also increase the reliability value (Hair et al. 2009). Factor analysis is applied to these multi-item scales as it is capable of identifying latent dimensions, and it is often used to measure attitudes. Before actually performing a factor analysis to bundle the multi-item scales, determining whether factor analysis is appropriate is another step. The Bartlett’s test of sphericity is a measure which determines the appropriateness of performing factor analysis. This test provides the statistical significance that the 1 c² p) -(1 * (p) * ² Z

(35)

correlation matrix has significant correlations among at least some of the variables (Hair et al. 2009). Another test of appropriateness is the Kaiser-Meyer-Olkin (KMO) measure, which measures whether the partial correlation among variables are small. Factor analysis is appropriate when the range of this measure is between 0.5 and 0.9 (Malhotra 2007).

Furthermore, to determine whether a multi-items scale can be grouped together in a single factor depends on their factor loadings. A factor loading represents the correlation between an original variable and its factor. When applying to the rule of thumb, loadings of 0.30 are considered significant for sample sizes of 350 or greater (Hair et al. 2009), which approximately equals the sample of this study. When a variable has no significant loadings, it may be eliminated from analysis depending on the variable’s overall contribution to the research. As a consequence, the factor model will be re-specified by deriving a new factor solution with those variables eliminated (Hair et al. 2009).

Team identification

As mentioned before, team identification signifies the extent to which a fan feels a psychological connection to a team, is involved and interested in the team, and sees the team as an extension of the self (Wann and Branscombe 1993). To assess this construct, a Dutch language version of the multi-item team identification scale is used. This scale is developed by Wann and Branscombe (1993), and is applied multiple times by other academics (Gwinner and Swanson 2003; Matsuoka, Chelladurai, and Harada 2003; Theodorakis et al. 2006; Theodorakis et al. 2009; Trail, Fink, and Anderson 2003; Wann, Bayens, and Diver 2004) since then. The Sport Spectator Identification Scale (SSIS) proved to have a strong reliability and validity in other settings (see Theodorakis et al. 2006; Wann & Branscombe 1993; Wann et al. 2001).

(36)

Kaiser-Meyer-Olkin (0.818) measure show acceptable values. Furthermore, as can be seen in table 4.1, all the loadings can be considered significant as they exceed the threshold value of 0.30.

Team Identification

Statements loadingsFactor

I think its important that [named team] wins 0.514 I strongly consider myself as a fan of [named team] 0.838 My friends see me as a big fan of [named team] 0.812 Being a fan of [named team] is very important to me 0.842 I strongly dislike the greatest rivals of [named team] 0.447 I often display the [named team] team’s name or insignia at my work, where I

live, and/or on my clothing 0.558 During the season I closely follow [named team] via ANY of the following

(television, radio, Internet or the paper) 0.730

Table 4.1 Factor analysis results for the multi-item scale ‘team identification’

Word-of-mouth

To asses the construct ‘word-of-mouth’ (WOM) among respondents, this study uses the opinion leadership scale of Godes and Mayzlin (2009). According to this article, “opinion leadership” is associated with a higher propensity to spread WOM among other consumers. As these behaviours might be beneficial for a football club, it also might be worthwhile to identify and consequently treat these supporters differently. The 7-point scale of Godes and Mayzlin (2009) proved to have adequate reliability in other settings.

(37)

addition, the factor model is re-specified without the inclusion of this statement. As the results in the right column of table 4.2 indicate, all of the factor loadings suffice to the minimal level of 0.30 (Hair et al. 2009).

Word-Of-Mouth Statements Factor loadings Factor loadings In general, I (1- rarely to 7- very often) like to talk to my friends and neighbours

about [named team] 0.775 0.771

Compared with my circle of friends, I am (1- rarely to 7- very often) asked

about [named team] 0.815 0.817

When I talk to my friends about [named team], I (1– give very little information

to 7– give a great deal of information) 0.843 0.847 During the past six months, I have told (1– no one to 7– a lot of people) about

[named team] 0.800 0.799

In discussions about [named team], (1– I usually tell my friends about [named

team] to 7– my friends usually tell me about [named team]) 0.189 -Overall, in my discussions with friends and neighbours about [named team], I

am (1 –not used as a source of advice to 7– often used as a source of advice) 0.779 0.784

Table 4.2 Factor analysis results for the multi-item scale ‘word-of-mouth’

Loyalty Intentions

In this study, loyalty intentions are derived from the intention that somebody has to attend future games. More specifically, this item is measured by the item “How likely are you to attend the [named team] games during the remainder of this season.” This 7 point scale was also implemented in the paper of Matsuoka, Chelladurai, and Harada (2003).

Residence

(38)

4.4 Analysis Methods

4.4.1 Regression techniques

To estimate the relationship between supporter demographics (independent variables) and team identification (dependent variable), multiple regression analysis is used. Simple regression is applied when the relationships between team identification (independent variable) and several supporter outcomes (dependent variable) are estimated. In order to derive the regression variate, which is a linear combination of the independent variables, ordinary least squares procedure (OLS) is applied (Hair et al. 2009). Furthermore, regression analyses is also able to capture moderating effects. This occurs when a second dependent variable changes the form of the relationship between another independent variable and dependent variable (Hair et al. 2009).

When validating the results of a regression analysis, several tests can be performed. The first major problem with regression models is multicollinearity. This refers to the situation that some independent variables tend to be highly correlated with another independent variable, resulting in unreliable parameter estimates (Leeflang et al. 2000). The variance inflation factor (VIF) accounts for multicollinearity and can be computed when performing a regression analysis. A common cut-off threshold is a VIF value of 10 (Hair et al. 2009). When validating a proposed model, one of the last stages is evaluating the residual diagnostics. The disturbances need to be normally distributed for the standard test statistics for hypothesis testing and confidence intervals to be applicable (Leeflang et al. 2000). To test whether the residuals follow a normal distribution there are two normality tests available, the Kolmogorov-Smirnov and the Shapiro Wilkinson test. In both cases the null hypothesis states that the sample comes from a normally distributed population.

(39)

observations, providing a single value for each observation. The D² measure divided by the number of variables involved (D²/df) is approximately distributed as a t-value. Observations having a D²/df value exceeding 4 in large samples (n>80) can be designated as possible outliers (Hair et al. 2009).

4.4.2 Segmentation techniques

In order to segment the fan base of a sport organization, several segmentation techniques are available nowadays, from which cluster analysis and latent class analysis are the most common ones. Latent Class Analysis (LCA) is a statistical method that classifies respondents into mutually exclusive groups with respect to a not directly observed (latent) trait. The LC analysis starts with the assumption that there is only one group, and subsequently estimates two, three, four, and finally n different classes, until a LC model is found that statistically fits the data best (Magidson and Vermunt 2004).

(40)

4.4.3 Mediation analysis

To derive the importance that sport marketing theory assigns to team identification, a mediation analysis is executed. Mediating variables lie in the causal sequence between an independent variable (X) and a dependent variable (Y) (Preacher and Hayes 2004). The most common application of mediation is to explain why a relationship between two constructs exist (Hair et al. 2009).

Mediation analysis involves assessing the effects of the mediator (M) in a relationship chain in which X is assumed to have a direct effect on M, which in turn causes changes in Y (Baron and Kenny 1986). Figure 4.1 depicts a basic mediation model, in which the common notations in the literature are used. Where c represent the direct effect of X on Y without considering M (Path 1), a represents the effect of X on M (Path 2), b represents the effect of M on Y adjusted for X, and c′ represents the direct effect of X on Y adjusted for M (Path 3). Thus, the indirect effect (i.e., the mediated effect) of X on Y through M is ab. Significance of the indirect effect ab provides evidence for mediation of the influence of X on Y by M (Zhang, Wedel, and Pieters 2009). If the mediating construct (M) completely explains the relationship between the two original constructs (eg. X and Y), then we term this complete mediation. But if some of the relationship between X and Y is not explained by the mediator, than we denote this as partial mediation (Hair et al. 2009).

Figure 4.1 basic mediation model

(41)

(42)

5 Results

This chapter kicks off with some descriptive results in section 5.1. Furthermore, as the study comprises several parts, this chapter is also divided according to these analyses. Section 5.2 discusses the conceptual model and hypotheses that are presented in chapter three. The cluster procedures which are executed in order to segment the database are presented in section 5.3. Finally, section 5.4 discusses the results of the mediation analysis that was executed to derive the importance of team identification.

5.1 Descriptive results

(43)

5.2 Conceptual model

The first three hypotheses all predict a relationship between a certain demographic variable and team identification. These hypotheses were tested by using multiple regression to examine simultaneously (1) whether the slope significantly differed from zero and (2) whether the slope was in the same direction as previous academic research suggested. The following regression equation was estimated:

e

3 3 2 2 1 1 0





b

X

b

X

b

X

Y

(1) Where

Y= Level of Team Identification

1

X = Gender of the respondent (0=male, 1=female)

2

X = Age

3

X = Distance (in kilometres) between home address and stadium ground

(44)

When a relationship is tested that only involves a single independent variable, simple regression is the proper statistical technique. In order to assess all attitude-behaviour links, three simple regression models are estimated. In addition, if the regression coefficient is found to be statistically significant, the value of the regression coefficient indicates the extent to which the independent variable is associated with the dependent variable (Hair et al. 2009).

The outcomes of the models are presented in table 5.1. First, the hypothesis stating that there is a positive relationship between the degree of team identification and relationship length is supported. This because the regression coefficient is both positive and significant. Second, as the simple regression coefficient modelling team identification and word-of-mouth is significant and positive, the hypothesis stating a positive relationship between these variables is supported. In addition, people with high levels of team identification are also more likely to engage in positive word-of-mouth. Third, a positive relationship between the degree of team identification and loyalty intentions can also be assumed, as the coefficient is both positive and significant. Fans with high levels of team identification also indicate that they have the intention to stay loyal to their favourite club. In addition, this hypothesis is also supported.

Furthermore, to test the moderating influence of perceived quality on each of the attitude-behaviour links, an moderator regression analysis is performed. This to determine whether a significant interaction effect existed between team identification and perceived quality, with the three different types of fan behaviours as dependent variables. The following regression equation was estimated:

e

3 2 1 0





b

X

b

Z

b

XZ

Y

(2) Where

Y= Fan behaviour outcome (Relationship length, Word-of-Mouth, or Loyalty intentions) X = Level of team identification

(45)

The results of the moderator regression analysis are presented in table 5.1. From these results it can be concluded that perceived value only has a significant moderating effect on the team identification – word-of-mouth link. In support of H7, we find that team identification with perceived quality has a significant and positive effect on fan behaviour solely in terms of engagement in word-of-mouth. In addition, we can state that fans with high levels of team identification are even more willing to express favourable word-of-mouth when it is accompanied by high levels of perceived quality.

Dependent variable Independent variable(s)

Model

fit Beta Sig. VIF R² Conclusion

Team identification 0,000 0,085

Gender -0,139 0,010 1 H1 rejected

Age -0,259 0,000 1 H2 rejected

Distance between home address and

stadium ground 0,012 0,820 1 H3 not supported

Relationship Length

Main effects-only model 0,003 0,023

Team identification 0,151 0,008 1 H4 supported

Moderator effects model 0,050 0,026

Team identification -0,151 0,711 50 Perceived quality -0,194 0,491 23

Team identification*Perceived quality 0,364 0,448 69 H7 not supported Word-Of-Mouth

Team identification 0,110 0,737 49 Perceived quality 0,449 0,051 24

Team identification*Perceived quality 0,806 0,038 68 H7 supported Loyalty Intentions

Team identification 0,079 0,840 49 Perceived quality -0,095 0,729 24

Team identification*Perceived quality 0,176 0,717 68 H7 not supported

(46)

However, when validating the moderator regression analysis we observe some multicollinearity issues. This because the VIF values in all three moderator effects models exceeds the threshold value of 10. After further examination of the bivariate correlation and the variables themselves, it shows that both team identification (0.806) and perceived quality (0.532) show high and significant correlations with Team Identification*Perceived quality. This is caused by the lack of variance that perceived quality has in this research setting. Which, in addition, results in highly correlated variables in the moderator regression analysis.

5.3 Cluster analysis

In order to develop appropriate value propositions for fans with different characteristics, attitudes, and behaviours, segmentation is a useful tool. Cluster analysis is a method used to derive different customer segments from a certain database. This study uses Latent Class Analysis (LCA) to classify respondents into mutually exclusive groups. There are several complementary approaches available for assessing the fit of LC models (Magidson and Vermunt 2004). The most commonly used measures are discussed in section 4.4 and summarized in table 5.2.

Classes Log Likelihood BIC AIC AIC3 AWE

1 -2858,7564 5757,69 5731,51 5738,51 5818,87 2 -2415,9274 4917,95 4861,85 4876,85 5086,98 3 -2296,6263 4725,27 4639,25 4662,25 4966,96 4 -2259,4758 4696,89 4580,95 4611,95 5115,16 5 -2206,4071 4636,67 4490,81 4529,81 5126,69

Table 5.2 fit statistics for different segment solutions

(47)

simplicity of the model into account (Banfield and Raftery 1993), which penalizes more drastically complex models.

Our results, presented in table 5.3, display a clear split among the supporter segments based on their demographics, attitudes, and behaviours. Cluster one, which comprises of 203 supporters (61,8%), shows average to high scores on team identification, word-of-mouth, and loyalty intentions. This group definitely is committed to the sport organization, as they deploy favourable behaviours and attitudes. In addition, when looking at the demographics, it can be stated that this cluster is the youngest (40 years) on average, and also has been attached to the organization for the shortest period (5 years) on average. Furthermore, these supporters reside in the most distant locations, as the distance from their home address to the stadium ground is 47 kilometres on average.

Cluster two scores the highest values on each of the attitude and behaviour variables. In addition, we can state that this specific sport organization has the biggest impact on the lives of these supporters compared with the other clusters. This is also indicated by the relationship length of this cluster, which lasts for 18,65 years on average. Furthermore, this cluster also has the highest intention to continue this relationship, as loyalty intentions are the highest for these supporters. Moreover, these 90 supporters represent the oldest segment with an average age of 49, and comprehend the highest percentage of males. Finally, these supporters live in the near proximity of the stadium, as the distance they need to travel in order to visit the stadium is lowest (30,3 kilometres) on average.

Cluster 1 Cluster 2 Cluster 3

61,80% 27,47% 10,73% Relationship Length 5,05 18,65 5,57 Team Identification 4,91 5,14 4,32 Word-of-mouth 4,76 4,97 4,16 Loyalty Intentions 6,70 6,75 2,22 Age 40,13 48,98 46,90 % Males 86,32% 93,17% 85,41% % Females 13,68% 6,83% 14,59% Distance from home address to stadium 47,032 30,3373 40,0673

Team identification as a key variable to distinguish supporters of a football club fan base